How DBAs Can Transform with AI: Insights from DTCC 2026

On August 22, 2026 the 17th China Database Technology Conference (DTCC 2026) concluded, featuring over two main venues and 16+ technical sessions covering database kernel innovation, cloud‑native and distributed practice, AI‑enabled data, vector databases, real‑time warehouses, and industry use cases.

DBA Director Zhang Wenjing of Lion Bridge Financing Lease (China) shared his practical experience of DBA‑AI transformation. He explained that traditional DBA work is often invisible until a failure occurs, and the AI wave creates anxiety about job replacement. Instead of abandoning DBA skills, he advocates amplifying existing capabilities with AI, leveraging automation of routine tasks (inspection, ticket handling, alert processing) to free time for AI‑driven work, and using existing servers, GPUs, and massive operational data as AI “fuel”.

He identified five advantages that enable DBA‑AI transition: abundant time, resources, data, scenarios, and experience. He outlined a five‑stage evolution: (1) adopt AI tools, (2) build AI assistants, (3) implement AI automation, (4) deploy AI agents, and (5) construct an AI platform that lets DBAs move from using AI to having AI autonomously complete work.

Bank technology expert Chen Zheng highlighted the bottlenecks of traditional big‑data architectures—data silos, report‑driven workflows, reliance on expert judgment, and complex component maintenance—and argued that lightweight “Big Data + AI” integration is essential for unlocking business value.

The proposed lightweight architecture consists of four layers—data, compute, intelligence, and application. The data layer provides a data lake and unified governance; the compute layer uses Flink for unified batch‑stream processing; the intelligence layer builds a full MLOps pipeline (feature engineering, model training, deployment, serving); the application layer delivers intelligent business decisions. AI is positioned as the “fuel” for the data layer and the “engine” for the intelligence layer, emphasizing collaboration rather than replacement.

From the perspective of open‑source pioneer Dai Lidong (White Whale Open‑Source), the next decade will shift from “how to let humans operate tools” to “how agents execute tasks within context and safety”. He introduced the “Agentic Data Stack” with five components: Execution Runtime, Data Engineering Harness, Semantic & Knowledge, Agentic Orchestration Control Plane, and Business Intent. The stack transforms AI generation into controlled delivery through a closed loop of Intent → Context → Plan → Skill → Execute → Validate → Review → Feedback.

Dai also warned that the scarcity in the Agentic AI era is not model generation but safe, production‑grade delivery. He recommends starting AI adoption with low‑risk, high‑frequency tasks such as data discovery, SQL draft generation, and log‑based diagnostic suggestions, then gradually moving to controlled execution and governance‑empowered automation for higher‑impact operations.

Both speakers concluded that the future of data engineering is not a fully unmanned system but a human‑defined goal, agent‑executed task, and harness‑constrained delivery model, positioning DBAs and data engineers as orchestrators of AI‑augmented pipelines.